Chemosphere 93 (2013) 1997–2003

Contents lists available at SciVerse ScienceDirect

Chemosphere

journal homepage: www.elsevier .com/locate /chemosphere

Fractal structures of single-walled carbon nanotubes in biologicallyrelevant conditions: Role of chirality vs. media conditions

0045-6535/$ - see front matter Published by Elsevier Ltd.http://dx.doi.org/10.1016/j.chemosphere.2013.07.019

⇑ Corresponding author. Tel.: +1 (803) 777 2288.E-mail address: salehn@engr.sc.edu (N.B. Saleh).

Iftheker A. Khan a, Nirupam Aich a, A.R.M. Nabiul Afrooz a, Joseph R.V. Flora a, P. Ariette Schierz b,P. Lee Ferguson c, Tara Sabo-Attwood d, Navid B. Saleh a,⇑a Department of Civil and Environmental Engineering, University of South Carolina, Columbia, SC 29208, USAb Department of Civil, Architectural, and Environmental Engineering, University of Texas, Austin, TX 78712, USAc Department of Civil and Environmental Engineering, Duke University, Durham, NC 27708, USAd Department of Environmental and Global Health, University of Florida, Gainesville, FL 32610, USA

h i g h l i g h t s

� SWNTs form chirality dependentfractal aggregates.� The (6, 5) enrichment is likely to

attain loose structures compared to(7, 6) ones.� Higher van der Waals energy is

responsible for tighter aggregates in(7, 6).� Presence of di-valent Ca2+ and FBS/

BSA lowers the Df values.� Steric hindrance likely caused the

loose aggregate structures.

g r a p h i c a l a b s t r a c t

a r t i c l e i n f o

Article history:Received 19 March 2013Received in revised form 2 July 2013Accepted 4 July 2013Available online 3 August 2013

Keywords:Fractal dimensionAggregationSingle-walled carbon nanotubesStatic light scatteringChirality

a b s t r a c t

Aggregate structure of covalently functionalized chiral specific semiconducting single-walled carbonnanotubes (SWNTs) was systematically studied employing static light scattering (SLS). Fractaldimensions (Df) of two specific chirality SWNTs—SG65 and SG76 with (6, 5) and (7, 6) chiral enrich-ments—were measured under four biological exposure media conditions, namely: Dulbecco’s ModifiedEagle Medium (DMEM), Minimum Essential Medium (MEM), Roswell Park Memorial Institute (RPMI)1640 medium, and 0.9% saline solution. The SWNTs exhibited chiral dependence on Df with SG65showing more fractal or loosely bound aggregate structures, i.e., lower Df values (range of 2.24 ± 0.03to 2.64 ± 0.05), compared to the SG76 sample (range of 2.58 ± 0.13 to 2.90 ± 0.08). All the Df valuesreported are highly reproducible, measured from multiple SLS runs and estimated with ‘random block-effects’ statistical analysis that yielded all p values to be <0.001. The key mechanism for such differencein Df between the SWNT samples was identified as the difference in van der Waals (VDW) interactionenergies of these samples, where higher VDW of SG76 resulted in tighter packing density. Effect of med-ium type showed lower sensitivity; however, presence of di-valent cations (Ca2+) in DMEM and MEMmedia resulted in relatively loose or more fractal aggregates. Moreover, presence of fetal bovine serum(FBS) and bovine serum albumin (BSA), used to mimic the in vitro cell culture condition, reduced theDf values, i.e., created more fractal structures. Steric hindrance to aggregation was identified as the keymechanism for creating the fractal structures. Also, increase in FBS concentration from 1% to 10% resultedin increasingly lower Df values.

Published by Elsevier Ltd.

1998 I.A. Khan et al. / Chemosphere 93 (2013) 1997–2003

1. Introduction

Single-walled carbon nanotubes (SWNTs), tubular structures ofsp3 hybridized carbon atoms, provide unique electrical, mechani-cal, and optical properties (Iijima and Ichihashi, 1993; Forro et al.2000; Baughman et al., 2002; Xia et al., 2003). Variations of tubediameter and atomic arrangement of carbon, together known aschirality, introduce added advantages to these helicoids and thuswiden the application premise to drug delivery, therapeutics, andelectronics (Thostenson et al., 2001; Bachilo et al., 2002; Minotet al., 2003; Chen et al., 2010). The growing market transfer of suchSWNT-laden applications has contributed to increased marketshare, projected to be $1.5 billion by 2015 (Mehta, 2010; NRC,2012). Such commercialization has generated added concerns forthis emerging nanomaterial and necessitates reliable studies forfate, exposure, and risk implications (Oberdorster et al., 2005; Kla-ine et al., 2008). Aggregation propensity of nanomaterials, particu-larly that of SWNTs, is well known in the literature to haveprofound implications for fate, transport, and biological exposure(Jaisi and Elimelech, 2009; Kang et al., 2009; Saleh et al., 2010).Particles at the nanoscale exhibit unique properties that allow sur-face energies to dominate their behavior. The interplay betweeninherent van der Waals forces and other specific and non-specificsurface interactions—e.g., electrostatic, hydrophobic, and ligand-receptor interactions, entropic forces, etc.—determine particlestability and organization at this scale (Elimelech et al., 1995;Espinasse et al., 2007). Aggregate structural organization can playa key role in nanomaterial aqueous fate as well as biological re-sponses. However, the state-of-the-art literature is limited toaggregate size evaluation with little or no consideration for theaggregate structures of the SWNTs.

Interfacial interactions of nanoscale materials get altered withthe change in surface chemistry and physical attributes and arewell known to have significant impact on their aggregation behav-ior. SWNTs, all carbon tubules, are also known to have uniqueaggregation behavior as a function of the aforementioned parame-ters (Zhang et al., 2009; Saleh et al., 2010; Forney et al., 2011; Bou-chard et al., 2012; Khan et al., 2013). To-date, the aggregationbehavior of SWNTs is systematically studied as a function of cova-lent and non-covalent surface functionality and in presence of arange of background chemistry, i.e., inorganic salts and various bio-macromolecules (Saleh et al., 2010; Bouchard et al., 2012; Khanet al., 2013). The key findings in aggregation studies of SWNTs pri-marily determine their aggregation state at various regimes withan effort to delineate mechanisms in response to the surroundingchemistry. The physicochemical differences thus contributing toSWNT cluster size are proved to have notable impact on their envi-ronmental fate and biological interaction. The tubular physical fea-tures, highly bundled aggregation state, as well as ability to adsorbhydrophobic biomacromolecules are known to have influencedbiological responses of SWNTs (Yang et al., 2010; Philbrook et al.,2011; Pasquini et al., 2012). It is also important to note that differ-ences in physicochemical attributes not only influence aggregationdynamics but also will influence particle–particle interaction andthus packing density of aggregated particles. Recently, covalentfunctionalization of SWNTs with varied aggregate structureshowed fractal structure dependent cytotoxicity (Pasquini et al.,2012); which generates an urgent need for systematic evaluationof aggregate structure in biological exposure conditions.

Structural morphology of aggregates has classically been ana-lyzed in colloidal literature by measuring fractal dimension (Df)(Weitz et al., 1985; Sorensen, 2001; Bushell et al., 2002). Analyticalmethods are established employing static light scattering (SLS) orlow angle X-ray scattering (LAXS) on polystyrene latex, silica,and other uniform colloids (Guo et al., 1990; Odriozola et al.,

2001; Bushell et al., 2002; Martinez-Pedrero et al., 2005). The foun-dation of such analytical techniques is based on the principle of an-gle-dependent scattering of electromagnetic waves (both visibleand X-ray waves); where the angular dependence is proved to bea function of aggregate packing density (Asnaghi et al., 1995;Sorensen, 2001; Bushell et al., 2002). Though the fundamentals ofDf determination have strong foundation on physical theories, yethave limited applications in nanoscale materials reflected throughonly a handful of studies reporting Df values (Chen et al., 2004; Sunet al., 2004; Kim et al., 2008; Lebedev et al., 2010; Kumar et al.,2012; Pasquini et al., 2012; Meng et al., 2013). Though angle-dependent detection of scattered visible waves (for SLS) is provedto be the most promising technique for Df measurement of nanom-aterials, significant challenges remain in obtaining reliable struc-tural information (Chen et al., 2004; Pasquini et al., 2012; Menget al., 2013). SWNTs, possessing a complex bundled and clusteredstructure, are uniquely different compared to classical colloidalmatrices; thus present key challenges of reproducibility for themeasured Df, using conventional SLS studies and data analysis.The studies to-date, presenting Df of SWNTs, fail to report confi-dence in their measured values by accounting for run-specificuncertainty and thus allow a sustained data gap for reliable Df pre-diction. Additionally, physical parameters, e.g., chirality, thoughinfluence optical, electronic, mechanical or chemical properties ofSWNTs, the impact of such parameters on aggregate structure isyet to be determined.

The objective of this study is to systematically address this datagap of reliable determination of Df, for chiral-specific SWNTs in dif-ferent biologically relevant exposure conditions. The chiral-specificSWNTs are covalently functionalized to achieve near-equal surfacefunctionality. Four cell culture media, i.e., Dulbecco’s Modified Ea-gle Medium (DMEM), Minimum Essential Medium (MEM), RoswellPark Memorial Institute (RPMI) Medium 1640, and 0.9% Salinesolution are considered as control media. Two different serum sup-plements, i.e., fetal bovine serum (FBS) and bovine serum albumin(BSA) were also added into the control media to mimic the in vitrocell culture condition. The scattered light intensity data was col-lected as a function of scattering angle and classical SLS techniquecoupled with systematic statistical analysis is employed to evalu-ate SWNT fractal dimension with (6, 5) and (7, 6) SWNT chiralities.Detailed physicochemical and electrokinetic characterizations areperformed using transmission electron microscopy (TEM), Ramanspectroscopy, and electrophoretic mobility (EPM) measurements.

2. Materials and methods

2.1. Chiral SWNTs

CoMoCAT SWNTs were procured from SouthWest NanoTech-nologies Inc. (SWeNT, Norman, OK, USA). Two different types ofsemiconducting SWNTs were tested in this study: SG65 (lot No.000-0031) and SG76 (lot No. 0020) tubes with predominantly (6,5) and (7, 6) chiral enrichment (Khan et al., 2013). The chirality dis-tribution of SWNTs was confirmed by NIR spectroscopy. Detailedchirality map and its NIR characterization data is reported in theearlier study (Khan et al., 2013). A chirality map showing (6, 5)and (7, 6) chiral indices is shown in Fig. S1.

2.2. Media

Biological osmotic fluid 0.9% saline solution (Cat. No. 101448-952, VWR) and three cell culture media, namely: DMEM (Cat. No.11885-084, invitrogen), MEM (Cat. No. 11575-032, invitrogen),and RPMI 1640 medium (Cat. No. 11835-030, invitrogen) wereused as stock media. The standard bovine serum albumin (BSA)

I.A. Khan et al. / Chemosphere 93 (2013) 1997–2003 1999

solution at 2 mg mL�1 concentration (Cat. No. 23210, Fisher Sci.)and heat inactivated fetal bovine serum (FBS) at similar concentra-tion (Cat. No. 16140-063, Invitrogen) were used to mimic the ser-um supplement conditions for in vitro cell culture systems.

2.3. Optimized Covalent Functionalization

The SWNTs were covalently functionalized following a two-stepprocess. 5 mg of SWNTs were bath sonicated (Branson 1510) in2.5 mL of 6 M HNO3 to remove metal impurities for a duration of1 h (Li et al., 2004). Later, the purified SWNTs were bath sonicatedwith 2.5 mL concentrated H2SO4/HNO3 mixture (3:1, 98% and 70%strength, respectively) for 4 h (Liu et al., 1998; Li et al., 2004). Aftereach step, SWNT suspension was filtered through 200 nm polypro-pylene (GHP) membranes (Pall Life Science, Port Washington, NY)and rinsed with Milli-Q water until the pH of the filtrate reached�6.5. The filtered sample was then dried for 3 h at 60 �C and keptin a silica gel chamber for 3 d. The conditions, i.e., sonication inten-sity, duration, oxidant composition, etc., were optimized using pre-viously established sonication technique (Khan et al., 2013) toachieve near-equal surface functionalization as verified by parallelcharacterization with XPS, Raman spectroscopy, and electropho-retic mobility measurements (EPM). Careful covalent functionali-zation of SG65 and SG76 yielded similar length tubes.

2.4. Characterization of SWNTs

Transmission electron microscopic characterizations of the sin-gle walled carbon nanotubes were performed to investigate theirstructural and bundling morphology in pristine, functionalized,and in-media forms. Details of TEM method are described else-where (Saleh et al., 2008, 2010). In short, 5 mL of �15 mg L�1

SWNT solution was prepared in ethyl alcohol (JT Baker, ACS Grade)and sonicated for 10 min using a sonic dismembrator (S-4000,Misonix). Suspensions were then placed in a bath sonicator for15 min prior to imaging and a drop of each suspension was placedon a 200-mesh copper TEM grid coated with carbon formvar (TedPella Inc.). Excess solvent was removed by wicking with filter pa-per and the grid was allowed to dry in a hot plate at 75 �C for15 min. Images were collected using an H-9500 (300 kV) TEM (Hit-achi High Technologies America, Inc, CA) with 0.18 nm resolution.Malvern Zetasizer (ZEN 3660) system was used to measure theEPM of the different chiral samples in various media conditionsusing established protocol described elsewhere (Saleh et al.,2008, 2010; Afrooz et al., 2013). In short, diluted stock solutionat �1 mg L�1 concentration was maintained. Medium solutionswere added immediately prior to the EPM measurements and allmeasurements were conducted with at least three replicate sam-ples to ensure reproducibility.

2.5. Fractal Dimension Measurement with SLS

Aggregate structure information was determined through frac-tal dimensions, Df measurements. ALV/CGS-3 precision goniometersystem (ALV-GmbH, Langen, Germany) was used in angle-depen-dent static light scattering (SLS) mode, which employs a 22 mWHe–Ne laser (632.8 nm wavelength) equipped with ALV/LSE-5004Light Scattering Digital Correlator (ALV-GmbH, Langen, Germany).For these experiments, 2 mL of solution of nanotubes in mediawere added to disposable borosilicate glass vials (Fisher Scientific,Pittsburg, PA). These vials were soaked in 2% cleaning solution (Ex-tran MA01, EMD Chemicals, Gibbstown, NJ), and thoroughly rinsedwith deionized water, and finally, oven dried (Fisher Scientific,Pittsburg, PA) under dust free conditions.

The scattered light intensity ‘‘I’’ was detected by a photoncounting module operating at 1.2 amperes and 5 volts made by

Perkin Elmer (Dumberry, Canada) and the data was collected as afunction of scattering angle that varied from 12� to 150� in 0.5�increment. To ensure that the suspensions maintain quasi-equilib-rium in terms of their colloidal stability, each measurement wasstarted 60 min after the addition of the SWNTs to media (Kimand Berg, 2000; Bushell et al., 2002). This time period was set byperforming time resolved dynamic light scattering experiments,where stable aggregate size achieving time guided such selection.The scattered light intensity (I) is related to the absolute value ofscattering vector (q), which is a function of the angle at whichthe intensity was collected at; Eq. (1) describes the relation be-tween I, q and Df.

I ¼ q�Df ð1Þ

Scattering vector q was determined using Eq. (2) that correlatesit with medium refractive index (n), laser wavelength (k) and scat-tering angle (h). Linear fractal regime of scattering was identifiedand fitted with a straight line to determine Df.

q ¼ 4pnk

sinh2

� �ð2Þ

Usually, the reported Df in the literature does not incorporaterun specific uncertainty; i.e., each angle-dependent run results inuncertainty in linear fit to acquire a Df value, which cannot be aver-aged to obtain a statistically reliable parameter. This paper ad-dresses run-specific uncertainty of I vs. q dataset utilizing aregression model with ‘random block-effects’. Statistical AnalyticalSystem (SAS) software package (Version 9.3, SAS Institute Inc.) wasemployed, where ‘proc mixed’ process—that is based on maximumlikelihood principle—with a random block variable captured run-specific uncertainty. All the Df values reported in this study hadundergone the statistical optimization. It is to be noted that Df val-ues for all conditions tested in this study showed high likelihoodshown by the strong confidence in the determined p-value<0.001. The software code used for the statistical analysis is in-cluded in Fig. S2.

3. Results and discussion

3.1. SWNT physicochemical characterization

The SWNT characterization with TEM and EPM measurementscontributed to the understanding of their morphological character-istics and electrokinetic behavior in different media conditions.Representative TEM micrographs of pristine and covalent function-alized SWNTs are presented in Fig. S3. Bundled and relatively longSWNT tubes are shown in Fig. S3(a and b), while substantial short-ening and debundling of tubes are evident from Fig. S3(c and d).Both the SG65 and SG76 chiral SWNTs showed similar debundlingand shortening that are similar to earlier reported work from theauthors (Khan et al., 2013). Also, such changes in SWNT morphol-ogy upon covalent functionalization are consistently observedelsewhere in the literature (Sun et al., 2002; Saleh et al., 2010).

Electrokinetic characterization of SWNTs shows similar nega-tive surface potential for SG65 and SG76 at different media condi-tions (Fig. 1). For SG65 and SG76, the EPM values were determinedto be �(1.34 ± 0.33) � 10�8 and �(1.81 ± 0.36) � 10�8 for DMEM,�(1.36 ± 0.4) � 10�8 and �(1.69 ± 0.31) � 10�8 for MEM,�(0.93 ± 0.3) � 10�8 and �(1.31 ± 0.35) � 10�8 for RPMI, and�(2.14 ± 0.46) � 10�8 and �(2.31 ± 0.40) � 10�8 m2 V�1 s�1 for sal-ine condition. The results show that chirality plays relatively insig-nificant role in electrokinetic properties of the SWNTs, whilecovalently functionalized with a strong oxidant. The similar valuesand relatively low magnitude of EPM are likely caused bysubstantial electrostatic screening due to the presence of high

DMEM MEM RPMI Saline0.0

-0.5

-1.0

-1.5

-2.0

-2.5

-3.0EP

M (1

0-8 m

2 V-1 S

-1)

SG65 SG76

Fig. 1. Electrophoretic mobility (EPM) of SWNTs under four different media i.e.,DMEM, MEM, and RPMI and 0.9% Saline solution. At least three separateexperiments were performed for each condition and data presented here are meanof three independent experiments with one standard deviation. Measurementswere carried out at a pH of �6.5 and a temperature of 20 �C.

0.006 0.01 0.015 0.02 0.0255

10

100

300

DMEM MEM RPMI Saline

Scat

terin

g In

tens

ity, I

(kH

z)

q (nm-1)

(a)

0.006 0.01 0.015 0.02 0.0251

10

100

300

DMEM MEM RPMI Saline

Scat

terin

g In

tens

ity, I

(kH

z)

q (nm-1)

(b)

DMEM MEM RPMI Saline2.00

2.25

2.50

2.75

3.00

Frac

tal D

imen

sion

, Df

SG65 SG76

(c)

Fig. 2. Scattering intensity (I) data with change in scattering vector (q) under fourdifferent media conditions: DMEM, MEM, RPMI and 0.9% Saline solution for (a)SG65 and (b) SG76 SWNTs. All the Static light scattering (SLS) measurements werecarried out at least thrice at a pH of �6.5 and at 20 �C. (c) Fractal dimension (Df) ofSG65 and SG76 SWNTs, estimated from (a) and (b), respectively. All the reported Df

values showed p value <0.001.

2000 I.A. Khan et al. / Chemosphere 93 (2013) 1997–2003

concentration of cations in biological media and saline solutions.Similar EPM values were reported earlier for higher salt conditions.For similar acid functionalization EPM of �(1.22 ± 0.20) � 10�8 and�(0.92 ± 0.20) � 10�8 m2 V�1 s�1 for SG65 and�(0.99 ± 0.08) � 10�8 and �(0.72 ± 0.02) � 10�8 m2 V�1 s�1 wasreported for 100 and 200 mM NaCl conditions, respectively (Khanet al., 2013). Mechanochemically stabilized SWNTs showed EPMvalues of �0.46 � 10�8 m2 V�1 s�1 at 55 mM NaCl (Saleh et al.,2010). Whereas, non-covalently functionalized, i.e., SDBS modifiedSWNTs, also showed relatively low values of f-potential; i.e., �18and �15 mV, for 140 and 200 mM NaCl, respectively (Ju et al.,2012).

3.2. Fractal structure of SWNTS in different media

The scattering intensity data are plotted with respect to the an-gle-dependent scattering vector in Fig. 2 for both SG65 and SG76SWNTs at all media condition. The fractal structure of SWNTs asdetermined by SLS is presented in Fig. 2(c). Overall, the Df showedsmall difference between different background media conditions,however, demonstrate a strong chirality dependence. The SG65SWNTs consistently displayed lower Df values ranging from2.24 ± 0.03 to 2.64 ± 0.05, compared to SG76 SWNTs with rangeof 2.58 ± 0.13 to 2.90 ± 0.08. The lowest Df was observed for DMEMand MEM media where as the higher values were yielded fromusing RPMI and saline background conditions. It is to be noted thatthe relative value of Df signifies aggregate structure conformation;i.e., closer the Df value is to the value of 3.0, more compact is theaggregate structure, whereas a Df value of near 1.0 value signifieslooser aggregate structure (Mukhopadhyay et al., 2012; Schaeublinet al., 2012). The relatively higher Df values for saline and RPMImedia are likely due to the sole presence of mono-valent cations,compared to mono- and di-valent cation presence in case of DMEMand MEM media conditions (Weitz et al., 1985; Kim and Berg,2000; Odriozola et al., 2001; Meng et al., 2013).

It is also to be noted that for all the media conditions, presenceof high amount of salt—i.e., >130 mM Na+ for all media and 1–2 mM Ca2+ for DMEM and MEM—has presented favorable aggrega-tion regime for the SWNTs with a nearly complete electrostaticscreening. Thus, the key interaction forces dominating in this re-gion is most likely the tube-tube interaction forces and van derWaals attractive interaction. It is well known that higher electro-static screening, results in ‘sticky colloids’, which typically yieldsloosely bound aggregate structure; i.e., low Df values (Weitzet al., 1985; Kim and Berg, 2000; Odriozola et al., 2001; Meng

et al., 2013). The SG65 tubes with relatively lower van der Waalsinteraction thus displays relatively low Df values for DMEM andMEM media conditions.

On the other hand, a substantially high van der Waals interac-tion that exists between SG76 tubes (Khan et al., 2013) has likelydominated the aggregation process and thus resulted in relativelyhigher Df values; i.e., more compact aggregate structures. It is note-worthy that di-valent electrolyte played a significant role in forma-tion of the aggregation structures. Absence of di-valent Ca2+ forboth RPMI and saline conditions has caused the SWNTs to packmore closely and yield higher Df values, compared to the DMEMand MEM media conditions with 1–2 mM Ca2+ cations. A slightly

SG65 SG762.00

2.25

2.50

2.75

3.00

Frac

tal D

imen

sion

, Df

Saline Saline w/ BSA

Fig. 4. Effect of BSA on fractal dimension (Df) of SG65 and SG76 SWNTs in controlmedia (i.e. 0.9% Saline solution). All the reported Df values showed p value <0.001for at least triplicate runs.

I.A. Khan et al. / Chemosphere 93 (2013) 1997–2003 2001

higher degree of electrostatic screening has most likely caused theSWNTs to be ‘stickier’ in presence of di-valent cations those resultsin such differences.

The fractal structure and relatively compactness of SWNTaggregates are further observed through TEM imaging in presenceof background media conditions. Fig. 3 shows differences in aggre-gate compactness, primarily caused through the tube–tube inter-action differences between SG65 and SG76. Relatively lowerdegree of tube–tube interaction for SG65 resulted in loosely boundSWNTs (Fig. 3a and c) compared to closely packed SG76 (Fig. 3band d), likely caused from higher van der Waals interaction be-tween these tubes. The TEM images are a visual proof of the Df

measurements, showing relative differences in compactness.

3.3. Role of biomolecules on fractal structure

Nanomaterial exposure media often contains biomolecules suchas globular protein BSA (Elgrabli et al., 2007; Qiu et al., 2010; Salehet al., 2010) and a more complex enzymetic and protein mixture ofFBS (Hu et al., 2011) at different concentrations. Understanding theaggregate structure in presence of these essential biomolecularadditives will likely provide further insight onto SWNT aggregationand its impact to biological responses (Pasquini et al., 2012). Fig. 4shows the role of BSA on Df. Addition of BSA shows a decrease in Df

for both SG65 and SG76 from 2.61 ± 0.13 to 2.52 ± 0.14 and2.89 ± 0.08 to 2.56 ± 0.06, respectively. These results indicate thatpresence of globular BSA proteins has likely caused a more looselypacked aggregate structure for both SWNTs and essentially dimin-ished the role of chirality on packing density. The likely mechanismfor such decrease in Df can be examined studying the interplay be-tween electrokinetic and physical contribution of BSA moleculeson aggregation.

Fig. 3. TEM micrographs of representative SWNTs: (a) SG65 w/RPMI (Df = 2.64 ± 0.05), (bw/Saline (Df = 2.89 ± 0.06).

Fig. S4a shows EPM values of both SG65 and SG76, where boththe tubes show lower electrostatic potential in presence of BSA. Itindicates that a slight decrease in EPM might have been caused byattachment of BSA molecules onto the SWNT surfaces and therebycausing a subsequent shielding of electrokinetic potential. Thus thelower electrokinetic potential contributed to increased ‘stickiness’of the tubes which has likely contributed toward lower Df values.Furthermore, the globular structure of BSA proteins has also likelycontributed to steric hindrances to attachment, and thereby causeda looser aggregate structure for both SG65 and SG76. Such evi-dences of steric interaction from BSA molecules were previouslyreported for SWNT aggregation (Saleh et al., 2010).

) SG76 w/RPMI (Df = 2.90 ± 0.08), (c) SG65 w/Saline (Df = 2.61 ± 0.13), and (d) SG76

2002 I.A. Khan et al. / Chemosphere 93 (2013) 1997–2003

A more complex biomolecule mixture FBS is also typicallyadded to exposure media during in vitro studies. The FBS mixtureessentially contains enzymes as well as the BSA proteins, whichhave high likelihood of causing substantial differences in aggregatestructure formation. Results presented in Fig. 5 shows consistentdecrease in Df for the increased addition of FBS (from 1% to 10%)for SG65 for all media conditions. For SG76, the trend was similarexcept for RPMI medium. It is to be noted that increased amount ofFBS did not reduce Df for SG76 in case of RPMI. We believe that theRPMI media lacks in di-valent cation presence, which reduced theelectrostatic screening and allowed for enhanced dominance of vander Waals interaction between SG76 tubes and resulted in compactaggregate structure. It is to be again noted that higher diameterSG76 possess higher van der Waals interaction forces when com-pared to SG65, as reported earlier (Khan et al., 2013).

Further evidences of loosening of SWNT aggregate structuresare observed through TEM imaging as presented in Fig. S5. In pres-ence of 10% FBS, both SG65 and SG76 showed looser packing ofaggregates; when compared to the media only cases. It is also evi-dent that in absence of FBS, the packing density is observed to behigher for SG76 compared to SG65, as consistently observed in ear-lier TEMs. The role of FBS on aggregate structure formation can alsobe evaluated using eletrokinetic properties and physical contribu-tion from biomolecular interaction. EPM results in presence of FBSshow a consistent decrease with the increase in FBS concentrationfor both SG65 and SG76 (Fig. S4b and c). The trend is similar to thedecrease in Df values as observed in Fig. 5. Similar to the BSA mol-ecules, a more complex mixture of biomolecules FBS has likelycaused sorption to SWNT surfaces and provided with strong shield-ing of electrokinetic interaction. The higher the FBS concentration,the higher is the FBS amount onto SWNTs and more is the shield-

DMEM MEM RPMI1.50

1.75

2.00

2.25

2.50

2.75

3.00 0% FBS 1% FBS 10% FBS

Frac

tal D

imen

sion

, Df

(a)

DMEM MEM RPMI1.50

1.75

2.00

2.25

2.50

2.75

3.00

Frac

tal D

imen

sion

, Df

0% FBS 1% FBS 10% FBS

(b)

Fig. 5. Fractal dimension (Df) values of (a) SG65 and (b) SG76 SWNTs in presence of0%, 1%, and 10% FBS added to DMEM, MEM, and RPMI 1640 medium. All reported Df

values showed p value <0.001 for triplicate runs.

ing. Moreover, increased amount of FBS also provided with highersteric hindrances between SWNTs, which also contributed towardlowering of Df values for both SG65 and SG76 SWNTs.

4. Conclusions

Fractal dimension can serve as a key parameter influencing bio-logical interaction and environmental fate of SWNTs. During bio-logical exposure, particularly for in vitro studies, the SWNTsundergoes interfacial interaction in response to the surroundingmedia conditions prior to interacting with the substrates; that usu-ally is placed at the bottom of the exposure cells. Thus, aggregationof SWNTs as a result of electrostatic screening alongside with un-ique fractal structure formation will influence the true exposure ofthe SWNT aggregates to the cellular entities. Lower the Df, looser isthe packing density, higher will be the fluid drag onto the SWNTclusters. Thus true exposure of low Df SWNT samples will be alot lower compared to that of more compact high Df aggregates.Similarly, fractal structure will also influence the fate of SWNTsin environmental systems. Fractal SWNTs will undergo similarfluid drag in surface water conditions that will likely result in amore persistent SWNT transport in such systems, compared tohigher Df more compact SWNT structures—which will likely settlefaster due to increased gravity and lower drag. It is important tonote that nanoparticulate recognition by cellular matrices dependon particle type and structure and thereby can be influenced byaggregate structure variation of the SWNTs. Moreover, SWNTtransport in groundwater systems will also be influenced by aggre-gate structure. The lower the Df, the more likely will be the propen-sity of the SWNT clusters to undergo physical filtration. ThusSWNTs with low Df in saline environment will likely undergo high-er degree of filtration. Therefore, aggregate structure information isan important physicochemical parameter that needs to be deter-mined for a complete and reliable understanding of nano-bio inter-action and nanomaterial fate in the environment.

Acknowledgements

Funding was provided by the National Science Foundation(CBET 0933484). We are grateful to Dr. Haijun Qian of ClemsonMicroscopy Center for his kind assistance in TEM imaging.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.chemosphere.2013.07.019.

References

Afrooz, A.R.M.N., Sivalapalan, S.T., Murphy, C.J., Hussain, S.M., Schlager, J.J., Saleh,N.B., 2013. Spheres vs. rods: the shape of gold nanoparticles influencesaggregation and deposition behavior. Chemosphere 91, 93–98.

Asnaghi, D., Carpineti, M., Giglio, M., Vailati, A., 1995. Light-scattering-studies ofaggregation phenomena. Phys. A 213, 148–158.

Bachilo, S.M., Strano, M.S., Kittrell, C., Hauge, R.H., Smalley, R.E., Weisman, R.B.,2002. Structure-assigned optical spectra of single-walled carbon nanotubes.Science 298, 2361–2366.

Baughman, R.H., Zakhidov, A.A., de Heer, W.A., 2002. Carbon nanotubes – the routetoward applications. Science 297, 787–792.

Bouchard, D., Zhang, W., Powell, T., Rattanaudompol, U.S., 2012. Aggregationkinetics and transport of single-walled carbon nanotubes at low surfactantconcentrations. Environ. Sci. Technol. 46, 4458–4465.

Bushell, G.C., Yan, Y.D., Woodfield, D., Raper, J., Amal, R., 2002. On techniques for themeasurement of the mass fractal dimension of aggregates. Adv. ColloidInterface Sci. 95, 1–50.

Chen, Q., Saltiel, C., Manickavasagam, S., Schadler, L.S., Siegel, R.W., Yang, H.C., 2004.Aggregation behavior of single-walled carbon nanotubes in dilute aqueoussuspension. J. Colloid Interface Sci. 280, 91–97.

I.A. Khan et al. / Chemosphere 93 (2013) 1997–2003 2003

Chen, K.J., Nair, N., Strano, M.S., Braatz, R.D., 2010. Identification of chirality-dependent adsorption kinetics in single-walled carbon nanotube reactionnetworks. J. Comput. Theor. Nanosci. 7, 2581–2585.

Elgrabli, D., Abella-Gallart, S., Aguerre-Chariol, O., Robidel, F., Rogerieux, F.,Boczkowski, J., Lacroix, G., 2007. Effect of BSA on carbon nanotube dispersionfor in vivo and in vitro studies. Nanotoxicology 1, 266–278.

Elimelech, M., Gregory, J., Jia, X., Williams, R.A., 1995. Particle Deposition andAggregation: Measurement, Modelling and Simulation. Elsevier.

Espinasse, B., Hotze, E.M., Weisner, M.R., 2007. Transport and retention ofcolloidal aggregates of C60 in porous media: effect of organicmacromolecules, ionic composition, and preparation method. Environ. Sci.Technol. 41, 7396–7402.

Forney, M.W., Anderson, J.S., Ameen, A.L., Poler, J.C., 2011. Aggregation kinetics ofsingle-walled carbon nanotubes in nonaqueous solvents: critical coagulationconcentrations and transient dispersion stability. J. Phys. Chem. C 115, 23267–23272.

Forro, L., Salvetat, J.-P., Bonard, R., Bacsa, Thomson, N.H., Garaj, S., Thien-Nga, L.,Gaal, R., Kulik, A., Ruzicka, B., Degiorgi, L., Bachtold, A., Schönenberger, C.,Pekker, S., Hernadi, K., 2000. Electronic and mechanical properties of carbonnanotubes. In: Tománek, D.E.R.J. (Ed.), Science and Application of Nanotubes.Kluwer Academic/Plenum Publishers, New York.

Guo, X.H., Zhao, N.M., Chen, S.H., Teixeira, J., 1990. Small-angle neutron-scatteringstudy of the structure of protein detergent complexes. Biopolymers 29, 335–346.

Hu, W.B., Peng, C., Lv, M., Li, X.M., Zhang, Y.J., Chen, N., Fan, C.H., Huang, Q., 2011.Protein corona-mediated mitigation of cytotoxicity of graphene oxide. ACSNano 5, 3693–3700.

Iijima, S., Ichihashi, T., 1993. Single-shell carbon nanotubes of 1-nm diameter.Nature 363, 603–605.

Jaisi, D.P., Elimelech, M., 2009. Single-walled carbon nanotubes exhibit limitedtransport in soil columns. Environ. Sci. Technol. 43, 9161–9166.

Ju, L., Zhang, W., Wang, X., Hu, J., Zhang, Y., 2012. Aggregation kinetics of SDBS-dispersed carbon nanotubes in different aqueous suspensions. Colloids Surf., A409, 159–166.

Kang, S., Mauter, M.S., Elimelech, M., 2009. Microbial cytotoxicity of carbon-basednanomaterials: implications for river water and wastewater effluent. Environ.Sci. Technol. 43, 2648–2653.

Khan, I.A., Afrooz, A.R.M.N., Flora, J.R.V., Schierz, P.A., Ferguson, P.L., Sabo-Attwood,T., Saleh, N.B., 2013. Chirality affects aggregation kinetics of single-walledcarbon nanotubes. Environ. Sci. Technol. 47, 1844–1852.

Kim, A.Y., Berg, J.C., 2000. Fractal aggregation: scaling of fractal dimension withstability ratio. Langmuir 16, 2101–2104.

Kim, T., Lee, C.H., Joo, S.W., Lee, K., 2008. Kinetics of gold nanoparticle aggregation:experiments and modeling. J. Colloid Interface Sci. 318, 238–243.

Klaine, S.J., Alvarez, P.J.J., Batley, G.E., Fernandes, T.F., Handy, R.D., Lyon, D.Y.,Mahendra, S., McLaughlin, M.J., Lead, J.R., 2008. Nanomaterials in theenvironment: behavior, fate, bioavailability, and effects. Environ. Toxicol.Chem. 27, 1825–1851.

Kumar, S., Aswal, V.K., Kohlbrecher, J., 2012. Size-dependent interaction of silicananoparticles with different surfactants in aqueous solution. Langmuir: ACS J.Surf. Colloids 28, 9288–9297.

Lebedev, V.T., Grushko, Y.S., Orlova, D.N., Kim, A.F., Kozlov, V.S., Sedov, V.P.,Kolesnik, S.G., Shamanin, V.V., Melenevskaya, E.Y., 2010. Aggregation inhydroxylated endohedral fullerene solutions. Fuller. Nanotub. CarbonNanostruct. 18, 422–426.

Li, Y., Zhang, X., Luo, Z., Huang, W., Cheng, J., Luo, Z., Li, T., Liu, F., Xu, G., Ke, X.,Li, L., Geise, H.J., 2004. Purification of CVD synthesized single-wall carbonnanotubes by different acid oxidation treatments. Nanotechnology 15, 1645–1649.

Liu, J., Rinzler, A.G., Dai, H., Hafner, J.H., Bradley, R.K., Boul, P.J., Lu, A., Iverson, T.,Shelimov, K., Huffman, C.B., Rodriguez-Macias, F., Shon, Y.-S., Lee, T.R., Colbert,D.T., Smalley, R.E., 1998. Fullerene pipes. Science 280, 1253–1256.

Martinez-Pedrero, F., Tirado-Miranda, M., Schmitt, A., Callejas-Fernandez, J., 2005.Aggregation of magnetic polystyrene particles: a light scattering study. ColloidSurf. A – Physicochem. Eng. Asp. 270, 317–322.

Mehta, M., 2010. 2009 NCMS Study of Nanotechnology in the U.S. ManufacturingIndustry. National Center for Manufacturing Sciences, Ann Arbor, MI.

Meng, Z., Hashmi, S.M., Elimelech, M., 2013. Aggregation rate and fractal dimensionof fullerene nanoparticles via simultaneous multiangle static and dynamic lightscattering measurement. J. Colloid Interface Sci. 392, 27–33.

Minot, E.D., Yaish, Y., Sazonova, V., Park, J.Y., Brink, M., McEuen, P.L., 2003. Tuningcarbon nanotube band gaps with strain. Phys. Rev. Lett., 90.

Mukhopadhyay, A., Grabinski, C., Afrooz, A.R.M.N., Saleh, N., Hussain, S., 2012. Effectof gold nanosphere surface chemistry on protein adsorption and cell uptakein vitro. Appl. Biochem. Biotechnol. 167, 327–337.

NRC, 2012. A Research Strategy for Environmental, Health, and Safety Aspects ofEngineered Nanomaterial. National Research Council of The NationalAcademies, Washington, DC.

Oberdorster, G., Oberdorster, E., Oberdorster, J., 2005. Nanotoxicology: an emergingdiscipline evolving from studies of ultrafine particles. Environ. Health Perspect.113, 823–839.

Odriozola, G., Tirado-Miranda, M., Schmitt, A., Lopez, F.M., Callejas-Fernandez, J.,Martinez-Garcia, R., Hidalgo-Alvarez, R., 2001. A light scattering study of thetransition region between diffusion- and reaction-limited cluster aggregation. J.Colloid Interface Sci. 240, 90–96.

Pasquini, L.M., Hashmi, S.M., Sommer, T.J., Elimelech, M., Zimmerman, J.B., 2012.Impact of surface functionalization on bacterial cytotoxicity of single-walledcarbon nanotubes. Environ. Sci. Technol. 46, 6297–6305.

Philbrook, N.A., Walker, V.K., Afrooz, A., Saleh, N.B., Winn, L.M., 2011. Investigatingthe effects of functionalized carbon nanotubes on reproduction anddevelopment in Drosophila melanogaster and CD-1 mice. Reprod. Toxicol. 32,442–448.

Qiu, Y., Liu, Y., Wang, L.M., Xu, L.G., Bai, R., Ji, Y.L., Wu, X.C., Zhao, Y.L., Li, Y.F., Chen,C.Y., 2010. Surface chemistry and aspect ratio mediated cellular uptake of Aunanorods. Biomaterials 31, 7606–7619.

Saleh, N.B., Pfefferle, L.D., Elimelech, M., 2008. Aggregation kinetics of multiwalledcarbon nanotubes in aquatic systems: measurements and environmentalimplications. Environ. Sci. Technol. 42, 7963–7969.

Saleh, N.B., Pfefferle, L.D., Elimelech, M., 2010. Influence of biomacromolecules andhumic acid on the aggregation kinetics of single-walled carbon nanotubes.Environ. Sci. Technol. 44, 2412–2418.

Schaeublin, N.M., Braydich-Stolle, L.K., Maurer, E.I., Park, K., MacCuspie, R.I., Afrooz,A.R.M.N., Vaia, R.A., Saleh, N.B., Hussain, S.M., 2012. Does shape matter?bioeffects of gold nanomaterials in a human skin cell model. Langmuir 28,3248–3258.

Sorensen, C.M., 2001. Light scattering by fractal aggregates: a review. Aerosol Sci.Technol. 35, 648–687.

Sun, Y.-P., Fu, K., Lin, Y., Huang, W., 2002. Functionalized carbon nanotubes:properties and applications. Acc. Chem. Res. 35, 1096–1104.

Sun, C.H., Li, F., Ying, Z., Liu, C., Cheng, H.M., 2004. Surface fractal dimension ofsingle-walled carbon nanotubes. Phys. Rev. B, 69.

Thostenson, E.T., Ren, Z.F., Chou, T.W., 2001. Advances in the science and technologyof carbon nanotubes and their composites: a review. Compos. Sci. Technol. 61,1899–1912.

Weitz, D.A., Huang, J.S., Lin, M.Y., Sung, J., 1985. Limits of the fractal dimension forirreversible kinetic aggregation of gold colloids. Phys. Rev. Lett. 54, 1416–1419.

Xia, Y.N., Yang, P.D., Sun, Y.G., Wu, Y.Y., Mayers, B., Gates, B., Yin, Y.D., Kim, F., Yan,Y.Q., 2003. One-dimensional nanostructures: synthesis, characterization, andapplications. Adv. Mater. 15, 353–389.

Yang, C.N., Mamouni, J., Tang, Y.A., Yang, L.J., 2010. Antimicrobial activity of single-walled carbon nanotubes: length effect. Langmuir 26, 16013–16019.

Zhang, S.J., Shao, T., Bekaroglu, S.S.K., Karanfil, T., 2009. The impacts of aggregationand surface chemistry of carbon nanotubes on the adsorption of syntheticorganic compounds. Environ. Sci. Technol. 43, 5719–5725.